Changing incidence of psychotic disorders among the young in Zurich

95 (2007) 9–18www.elsevier.com/locate/schres

Schizophrenia Research

Changing incidence of psychotic disorders among theyoung in Zurich

Vladeta Ajdacic-Gross a,⁎, Christoph Lauber a, Inge Warnke a,Helene Haker a, Robin M. Murray b, Wulf Rössler a

a Research Unit for Clinical and Social Psychiatry, Psychiatric University Hospital, Zurich, Switzerlandb Institute of Psychiatry, King's College London, London, United Kingdom

Received 3 April 2006; received in revised form 8 June 2007; accepted 10 June 2007Available online 16 July 2007

Abstract

There is controversy over whether the incidence rates of schizophrenia and psychotic disorders have changed in recent decades.To detect deviations from trends in incidence, we analysed admission data of patients with an ICD-8/9/10 diagnosis of psychoticdisorders in the Canton Zurich / Switzerland, for the period 1977–2005. The data was derived from the central psychiatric registerof the Canton Zurich. Ex-post forecasting with ARIMA (Autoregressive Integrated Moving Average) models was used to assessdepartures from existing trends. In addition, age-period-cohort analysis was applied to determine hidden birth cohort effects. Firstadmission rates of patients with psychotic disorders were constant in men and showed a downward trend in women. However, therates in the youngest age groups showed a strong increase in the second half of the 1990's. The trend reversal among the youngestage groups coincides with the increased use of cannabis among young Swiss in the 1990's.© 2007 Elsevier B.V. All rights reserved.

Keywords: Psychotic disorders; Schizophrenia; Incidence; Time series; Cannabis; Switzerland

1. Introduction

There is a growing body of evidence indicating thatincidence rates of schizophrenia vary more than one wouldexpect from an “egalitarian disorder” (McGrath, 2005). Forexample, there is considerable evidence of variations inincidence according to gender, urban or rural upbringing,and ethnicity (van Os, 2004). In recent decades, evidenceof a rate decrease (Brewin et al., 1997; Der et al., 1990;

⁎ Corresponding author. Research Unit for Clinical and SocialPsychiatry, Psychiatric University Hospital, Militärstr. 8, PO Box1930, CH-8021 Zürich, Switzerland. Tel.: +41 442967433; fax: +41442967449.

E-mail address: [email protected] (V. Ajdacic-Gross).

0920-9964/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.schres.2007.06.001

Eagles et al., 1988; Eagles and Whalley, 1985; Suvisaariet al., 1999; Takei et al., 1996; Waddington and Youssef,1994; Woogh, 2001), has been reported from studies ofincidence or first admission rates, the latter being a proxyfor incidence rates (Geddes et al., 1993; Jones et al., 1997).A few studies have reported constant incidence rates or firstadmission rates (Harrison et al., 1991;Oldehinkel andGiel,1995; Osby et al., 2001), and others have reportedincreasing rates (Boydell et al., 2003; Preti and Miotto,2000; Tsuchiya and Munk-Jorgensen, 2002).

We set out to examine the sex and age-specific trendsof first admission rates of people with psychotic disordersin the Canton Zurich/Switzerland from 1977–2005, todetect incidence trends as well as departures from trends.To determine the latter, two different approaches were

mailto:[email protected]

http://dx.doi.org/10.1016/j.schres.2007.06.001

Fig. 1. Selection of first admissions with F2 diagnoses from the centralpsychiatric register of the Canton Zurich, Switzerland.

10 V. Ajdacic-Gross et al. / Schizophrenia Research 95 (2007) 9–18

used: ex-post forecasting within ARIMA modelling, andage-period-cohort analysis.

2. Materials and methods

First admission data for schizophrenia and otherpsychoses was obtained from the Psychiatric CaseRegister in Zurich. All mental health services in theCanton Zurich provide detailed information aboutdiagnostic, treatment-related and socio-demographiccharacteristics of all patients to the central psychiatricregister. The data is collected by means of a basicdocumentation system, which assesses information viastandard forms completed both at admission, and atdischarge (Lauber et al., 2005).

The register contains basic information on allpsychiatric inpatients in the Canton Zurich since 1974.The place of residence has been recorded since 1977.The catchment area, the Canton of Zurich, is a mixedurban-rural area with a population of 1.2 million, whichcomprises about one sixth of the total Swiss population.

Discharge diagnoses were used. The dischargeprotocol allowed up to 4 diagnoses until 1997, and upto 9 diagnoses from 1998 onwards. The diagnoses werebased on ICD-8 from 1974–1978, on ICD-9 from 1979–1991, and on ICD-10 since 1992. Two different definitionsof schizophrenia/psychotic disorder were applied in thisanalysis: firstly, a narrow variant restricted to codes F20(ICD-10) and 295 (ICD-8 and ICD-9) and, secondly, abroader variant covering all F2 codes (ICD-10) and thecodes 295, 297, 298, 299 (ICD-8 and ICD-9). Theintroduction of ICD-10 caused a shift in the reporting ofnarrowly defined schizophrenia, which was compensatedby other F2-categories. Therefore, to avoid biasing effectsof changing definitions over time, our analyseswere basedon the broad definition of schizophrenia and otherpsychotic disorders.

Between 1977 and 2005, a total of over 180,000admissions were registered. The narrow schizophreniadiagnosis was applied to 38,000 admissions, and thebroader psychotic disorder diagnosis to more than52,000 admissions. Slightly more than 10% of therecords related to people not living in the Canton Zurich,and were therefore excluded.

The denominator (i.e. the population of CantonZurich) was interpolated from the census data (1970,1980, 1990, 2000).

2.1. Identification of patients at first admission

The admission records lacked for the most part aunique identification number for each patient. Thus,

several filtering and matching steps had to be applied inidentifying the first admissions of F2-patients. Theyincluded the following criteria (Fig. 1):

mention of F2-diagnosis on the discharge protocol
˙˙ exclusion of previous psychiatric admissions in theadmission protocol (self-reported); it was not possi-ble to differentiate between previous F2 admissionsand admissions due to other diagnoses; therefore, the

Table 1APC analyses of first admission data of patients with psychoticdisorders in the Canton Zurich, Switzerland, from 1977–2005; 5-yearperiod and age intervals (age range 15–49), by sex

Males FemalesEffects df a Deviance p-value df a Deviance p-value

1. Intercept 41 947.8 .000 41 244.2 .0002. Age 35 93.5 .000 35 132.5 .0003. Period 36 917.9 .000 36 160.0 .0004. Cohort 30 618.7 .000 30 173.4 .0005. Age period 30 63.1 .000 30 45.4 .0366. Age cohort 24 59.8 .000 24 20.5 .6697. Period cohort 25 342.1 .000 25 122.2 .0008. Age period

cohort b20 38.6 .008 20 18.0 .588

a df: degrees of freedom.b Model with 1 restricted parameter (set on the first age estimate).

11V. Ajdacic-Gross et al. / Schizophrenia Research 95 (2007) 9–18

F2 first admissions were underestimated to someextent

˙ exclusion of second (and later) occurrence of anidentification code after matching by sex and date ofbirth; the identification codes were specific for eachhospital, however, they turned out not to be uniquewhich meant that additional matching by sex anddate of birth was necessary

˙ inclusion of residents of the Canton Zurich; this stepwas introduced rather late because of pragmatic rea-sons (programming)

˙ exclusion of records with identical informationregarding sex, date of birth and the zip code; thisstep ruled out those patients who were treated inseveral hospitals, but failed to deliver any appropriateinformation about prior hospitalization

˙ exclusion of records with identical information onlyabout sex and date of birth; this step was morerestrictive than the previous one.

In this way, we identified 3972 first admissionswhich received a narrow schizophrenia diagnosis, and7230 first admissions with a broader psychotic disorderdiagnosis. In addition, we found 328 (837) duplicateentries with regard to sex and date of birth, which ismore than one might expect. Therefore we calculatedthe incidence figures without using duplicate entries.Only the first record of a sex and birth date match waskept. This measure contributed to a certain amount ofunderestimation of first admission rates.

2.2. ARIMA modelling

To ascertain departures from trends, the time serieswere modelled by means of ARIMA, also known as theBox–Jenkins approach (Box and Jenkins, 1970; Gottman,1981). Rates in 1977–1995, that is, the period before thepresumed increase of first admission rates, served to buildpreliminary models. The rates were calculated semi-annually to provide a greater number of observations.ARIMA models rely on estimation of autoregressive(AR) and moving average (MA) processes inherent in thetime series, eventually preceded by data transformationoperations and differencing (de-trending). These modelswere used to ex-post forecast values from 1996 onwardswhen the increase of the first admission rates in the youngbegan. The rest of the time series from 1996 onwardsserved to confirm the validity of the previously statedmodel. Deviations from the forecasted values (and their95% confidence intervals) indicate that specific fluctua-tions or trend reversals emerge, which do not conform tothe former time series patterns.

2.3. Age-period-cohort (APC) analysis

The main aim of APC analysis is to describe historicalchange by disentangling direct, immediate effects asagainst delayed birth cohort effects on any outcomevariables such as incidence or mortality of a disease. APCanalysis is usually based on repeatedly (for example,annually) collected age-stratified data pooled in so-calledcohort tables (Fienberg andMason, 1985). It differentiatessimultaneously between the effects of age (age effects),the effects of historical change (period effects), and thegenerational succession (cohort effects). The mainobstacle is that there is a redundancy between linearage, period and cohort effects. Any two of the dimensionsage, period and cohort determines the third. Thus,additional restrictions or conventions are required (Hol-ford, 1985, 1991). In this study, we used a simpleapproach to deal with the identification problem in APCanalysis. Analogous to the drift-approach (Clayton andSchifflers, 1987a,b), age effects act as a mandatorycomponent in all models, and, furthermore, as the targetfor subsequent constraints. Since age effects were shownto largely determine the data structure in preliminary one-factor and two-factor analyses (Table 1), the full APCmodels were calculated after restricting age estimatesaccording to the results in the preliminary AC models. Inthe analyses reported below, we confined the restrictionsto the first age estimate only, that is, by adopting therespective parameter from the AC model.

APC analyses were calculated within the frameworkof log-linear analysis, i.e., logit models, based on a cohorttable with 5-year intervals for age (15–19, 20–24,…,50–54) and period (1977–81, 1982–86 etc.; the lastperiod relies only on data for the four years (2002–5)).The goodness of fit or, rather, the lack of fit, was assessed

Fig. 2. Age-specific first admission rates for schizophrenia and other psychotic disorders in the Canton Zurich, Switzerland, 1977–2005; men, 3-yearmoving averages, ages 15–19, 20–24, 25–29, 30–34 and 35–39.


by the deviance or likelihood-ratio. The deviance of allone-or two-effect models was assessed routinely.

The values depicted in the figures represent estimatesafter exponentiation. All computations were performedusing SAS (SAS Institute Inc., 1996).

3. Results

Between 1977 and 2005, there were 8091 firstadmissions who received a diagnosis of schizophreniaor other psychoses at discharge: 3767 (46.5%) men and4324 women (53.5%). After excluding duplicate entrieswith regard to sex and the date of birth, there remained7230 first admissions, i.e., 3331 men and 3899 women.

Fig. 3. Age-specific first admission rates for schizophrenia and other psycho3-year moving averages, ages 15–19, 20–24, 25–29, 30–34 and 35–39.

The overall first admission rates in men were on asimilar level round the year 1980 and after 2000 (∼20per 100,000), whereas the rates of women decreasedfrom ∼30 to ∼20 per 100,000. The age-specific firstadmission rates (rates available on request) for menshowed a steep increase in young adulthood resulting ina peak in the age group 20–24. In women, the peak inthe rates occurred in the age-groups 25–29 and 30–34,and was lower than in males. However, the decreasethereafter was distinctly more gradual than in males. Theage shapes were similar over the whole period.

The longitudinal perspective shows a slightlydecreasing trend in first admission rates in most maleage groups, and a decrease in all female age groups until

tic disorders in the Canton Zurich, Switzerland, 1977–2005; women,

Fig. 4. Ex-post forecasting of semi-annual first admission rates for schizophrenia and other psychotic disorders of 15–19-year old men in the CantonZurich, Switzerland; rates (dark bold line), predicted/forecasted values (dark thin line) and 95% confidence intervals (grey lines); ARIMA modellingwas based on 1977–1995 values.


the 1990s. In the 1990s the trend changed for theyoungest male age group (15–19 years) and, with sometime delay, also in the 20–24 year olds (Fig. 2). Nosimilar change of trend is obvious in the youngestfemale age groups (Fig. 3). No trend reversal is apparentin older age groups. Over the entire time series, theyoung male age groups seem to be most prone tofluctuating patterns, and also show an interim peak inthe early 1980s.

The ARIMA analyses support this interpretation. Ex-post forecasting of the rates after 1995 showed that thevalues of the youngest male age groups (15–19 and 20–

Fig. 5. Ex-post forecasting of semi-annual first admission rates for schizophreZurich, Switzerland; rates (dark bold line), predicted/forecasted values (darkon 1977–1995 values.

24 year old) temporarily left the 95%-confidenceintervals (Figs. 4 and 5). This is despite the fact thatthe confidence intervals are broad since relatively fewobservations were used in the model building. In allother sex-age subgroups, the departures from trend wereless impressive or not existent (data available onrequest).

Table 1 shows the improvement of the model fit inAPC-analysis while introducing the A-, P- and C-effects. It is obvious from the deviance values that thefull APC model fitted the males data clearly moresatisfactorilythan any two effects model. In females, the

nia and other psychotic disorders of 20–24-year old men in the Cantonthin line) and 95% confidence intervals; ARIMA modelling was based

Fig. 6. Age-period-cohort analysis of male first admissions for schizophrenia and other psychotic disorders in the Canton Zurich, Switzerland, 1977–2005; logit analysis of 5⁎5-standard cohort tables, first age estimate being fixed.


full APC model provided only minimal improvementwith respect to the AC model. The estimates derivedfrom the APC analyses are depicted in Figs. 6 and 7. TheA-section shows the shape of the age-estimates, the P-section shows the period estimates and the C-section thecohort estimates. APC analysis of first admissions ofmale patients with schizophrenia and other psychoses(Fig. 6) indicates that the increase of the rates during the

Fig. 7. Age-period-cohort analysis of female first admissions for schizophre1977–2005; logit analysis of 5⁎5-standard cohort tables, first age estimate b

1980s is related to period effects, whereas the recentincrease of the rates in the young is partitioned intoperiod and cohort effects. The period estimates suggest atemporary increase whereas the cohort estimates suggesta more lasting phenomenon. Since the latter rely mainlyon the most recent birth cohorts with few observations,they also could be interpreted as short-term age-periodinteractions. In females, the birth cohorts estimates show

nia and other psychotic disorders in the Canton Zurich, Switzerland,eing fixed.


only a slight disturbance among the most recent cohorts(Fig. 7).

4. Discussion

Since the 1980s, possible changes in incidence rates ofschizophrenia have been discussed extensively. Whilesome researchers argued that there is a real decrease(Brewin et al., 1997; Eagles and Whalley, 1985; Geddeset al., 1993), others have suggested artificial effects(Allardyce et al., 2000; Boydell et al., 2003; Kendell et al.,1993; Munk-Jorgensen and Mortensen, 1992; Osby et al.,2001). An alternative perspective has come recently fromDanish data: after a decrease of schizophrenia incidencerates up to the 1980s (Munk-Jorgensen and Mortensen,1992), the trend reversed and the rates then began to rise(Tsuchiya and Munk-Jorgensen, 2002). An increase inschizophrenia incidence rates since the 1960s was alsodemonstrated in Camberwell (south-east London), partic-ularly among the young age groups (Boydell et al., 2003).

The present data shows that the initial stationary, orpossibly decreasing, trend in the Canton Zurich/Switzer-land has selectively changed among adolescents andyoung adults during recent years. Males seem to be moreaffected than females. The deviation from the previouspattern was demonstrated by two different methodicalapproaches, i.e., ex-post forecasting with ARIMAmodels, and age-period-cohort (APC) analysis.

The figures are based on first admissions of patientswith schizophrenia and other psychotic disorders topsychiatric inpatient services in a catchment area with1.1 million residents between 1977 and 2005. Sinceschizophrenia is a disease in which patients are usuallyadmitted to hospital (Geddes et al., 1993; Jones et al.,1997), first admission figures are only a proxy of the realincidence. In contrast to most other studies dealing withthe change of schizophrenia incidence rates, the Zurichregister offers both a large data base, and a relatively longobservation period. We compared different inclusioncriteria to account for the possibility that all of the dupli-cate entrieswith regard to sex and the date of birthwere re-admissions instead of first admissions. The results did notalter substantially. As in other studies (Oldehinkel andGiel, 1995), the decrease of the rates was not so markedwhen a broader definition of psychotic disorderswas used,thus suggesting some change occurring between diagnos-tic criteria over the recent decades. The use of the broaderdefinitionwas necessary because the diagnostic practice innarrowly defined schizophrenia obviously changed withthe introduction of the ICD-10 coding in 1992.

The limitations of this study are those of the registerdata. The figures are based on first admissions of psychotic

disorders in inpatient services and do not exactly mirrorthe time of disease onset. Moreover, the information onprevious admissions derived from the admission protocolturned out to be lacking or inappropriate in someinstances. Thus, we introduced an additional control byexcluding cases after matching by combined sex/birthdate/ZIP code data. Presumably, the matching was lesssuccessful at the beginning of the examined period, that isin the 1970s and the early 1980s, than later.

The register data does not include information aboutthe lifetime use of legal and illicit drugs. Moreover, thelimitations stem from effects, which might interfere withtime series of incidence data: change of classificationsystems (ICD-8, -9 and -10), change of diagnostic habits(mainly schizophrenia vs. other diagnoses of psychoticdisorders), changes in health service organisation, delaysin the use of inpatient services, early interventionprogrammes, and, finally, use of outpatient and alternativeservices. However, it is not plausible that such effectswould bias the incidence rates of schizophrenia and otherpsychoses in an age and sex-specific manner. Moreover,the increasing importance of outpatient services wouldsuggest a decrease rather than an increase of rates. Toexclude artificial effects in register-based incidence ratesis more problematic with regard to long-term change, thanto short-term fluctuations.

Provided that the limitations mentioned above do notbasically change any conclusions reached, the Zurichfigures have shed a new light on the incidence of psy-chotic disorders. In Zurich, as elsewhere, incidence ratesof schizophrenia and other psychoses have declined overrecent decades. As elsewhere, the decrease seems to haveaffected women rather than men (Geddes et al., 1993;Jones et al., 1997; Kendell et al., 1993; Waddington andYoussef, 1994). Specifically, in Zurich there has been atrend-reversal in incidence rates in young men. The ratesfor young men in former years have also exhibitedremarkable fluctuations— such as the peak in the 1980s.

There are not many plausible explanations for suchdeviating incidence patterns in the young (Maki et al.,2005). It is unlikely that such short-term fluctuations inincidence could be related to risk factors related topregnancy and delivery, such as famine experienced bythe mother during pregnancy, prenatal infections orobstetric complications. Such risk factors appertain tobirth cohort effects and determine slow change phenom-ena. Among the known risk factors, drug use is the onlyone which might generate fast changing incidences ofpsychotic disorders on the population level (McCormickand Goff, 1991).

In fact, during the 1990s the patterns of drug use havemarkedly changed among young people in Western


Europe. Two trends have been obvious. On the onehand, cannabis has become more popular and broadlyavailable, and on the other, a new party and dancingscene has emerged which extensively used ecstasy(MDMA) and to a lesser extent other syntheticamphetamines and hallucinogens (von Sydow et al.,2002). According to information from the Zurich police(personal communication from Dr. Bovens of theScientific Department), the latter drugs as well asphencyclidine (PCP) have remained clearly a periph-eral phenomenon in the region of Zurich. Cannabis andecstasy (or any interactions of these two substanceswith other ones) appear to be the most promisingcandidates to explain the trend change in incidence offirst admissions in psychotic disorders.

In recent years, cannabis use has been the focus ofmuch attention in the research on etiology of psychoticdisorders. The ability of cannabis in inducing an acutepsychosis has been known for a long time. Since the endof the 1980s there has been speculation about anassociation between cannabis and psychotic disorders,which goes beyond transient effects (Macleod et al.,2004). Evidence for cannabis as a risk factor has comefrom several longitudinal studies (Arseneault et al.,2002; Boydell et al., 2006; Fergusson et al., 2005, 2003;Henquet et al., 2005a; van Os et al., 2002), the mostimpressive being the Swedish conscript cohort study(Zammit and Lewis, 2004), which assessed admissionsfor schizophrenia in about 50,000 former militaryconscripts using data-linkage methods. They found adose-response relationship between former cannabis useand later onset of schizophrenia. The odds ratiosincreased from 1.2 (ever used cannabis before conscrip-tion) to greater than 6 in persons with frequent cannabisuse before conscription. In meta-analyses, the odds ratiofor schizophrenia related outcome in cannabis users wasshown to be slightly above 2 after adjustment for poten-tial confounders (Arseneault et al., 2004; Henquet et al.,2005b).

Despite growing evidence for an association betweencannabis use and the onset of schizophrenia, there is stillsome scepticism about the link between the two(Degenhardt et al., 2003; Hall et al., 2004). A majorissue is whether cannabis use is a potential cause ofpsychotic disorders or rather triggers the onset invulnerable persons (Caspi et al., 2005). The secondsource for scepticism has emerged from the fact that noupward trends in incidence rates of psychotic disordershave been shown in a population as a whole despite theincreasing use of cannabis products (Arseneault et al.,2004;Macleod et al., 2006; Rey and Tennant, 2002). TheZurich figures appear to provide the missing link.

The evidence for an association between cannabisuse, and the changing first admission trends in Zurich,includes the following:

˙ the increasing first admission rates of patients withschizophrenia / psychotic disorders in the second halfof the 1990's: cannabis availability (hemp shops) andconsumption distinctly increased in the 1990's; Swissdata from the Health Behaviour in School-AgedChildren (HBSC) survey shows that lifetime preva-lence of cannabis use in 15–16 year old teenagers rosefrom 15% (boys) / 5% (girls) in 1990 to 41% / 30% in1998 and to 50% / 39% in 2002 (Delgrande Jordanet al., 2004; Kuntsche, 2004)

˙ the increase tended to hit the youngest age groups:epidemiological data on the use of illicit drugs haveindicated that the popularity and frequency ofcannabis use in the 1990's has changed mostdistinctly in adolescents and young adults (DePreux et al., 2004; Drewe et al., 2004);

˙ the trend reversal started first in the teen years and,after some delay, in the twenties: this is in line with thefinding reported above that cannabis use precipitates theonset of schizophrenia and psychotic disorders

˙ the trend reversal was stronger in men than inwomen: while the lifetime prevalence rates ofcannabis use in men and in women differ onlyslightly, men were shown to consume distinctly morefrequently (see above: dose-response relationship);for example, in the 2002 SMASH survey 13% of 16–20 year old boys (4% of girls) indicated daily use ofcannabis (De Preux et al., 2004; Narring et al., 2004)

The ecstasy hypothesis in psychosis research hasyielded ambiguous evidence. Although there are casereports suggesting that persistent psychosis may occurafter an intake of ecstasy (Van Kampen and Katz, 2001),the evidence from epidemiological studies is lacking.Psychopathological symptoms seem to be associatedmainly with regular cannabis use in recreational ecstasyusers (Daumann et al., 2004).

It seems possible that interactions between cannabisand other amphetaminic or hallucinogenic drugs increasethe risk of psychotic disorders. A detailed assessment ofsubstance use patterns might provide further informationon this issue in future investigations.

To sum up, the change in incidence of psychoticdisorders depends on environmental influences, whichmay exhibit systematic short-term fluctuations. Suchchanges were demonstrated in this analysis of the Zurichdata, for the period 1977–2005, using two differentstatistical approaches. These changes affect mainly


young age groups, and men rather than women. Theypresumably mirror the changes in use of cannabis sincethe 1990's. In the 2000's the views on cannabis havebegun to change. The availability of cannabis productshas been reduced (for example, by the closing of hempshops, and more rigorous controls by the police), and thelegalization of the sale of cannabis products with highTHC content has been disapproved by parliament.

Role of the funding sourceNone.

ContributorsAll authors listed below have contributed to this work either in data

preparation and analysis (I. Warnke, C. Lauber, V. Ajdacic-Gross), inwriting (V. Ajdacic-Gross, C. Lauber, W. Rössler) and in discussionand reformulation of the paper (C. Lauber, H. Haker, R. Murray, W.Rössler). All authors have reviewed the manuscript and approved itscontents.

Conflicts of interest
None.
Acknowledgements

We were thankful for the comments of Michael Schaub and HansWydler.

References

Allardyce, J., Morrison, G., Van Os, J., Kelly, J., Murray, R.M.,McCreadie, R.G., 2000. Schizophrenia is not disappearing insouth-west Scotland. Br. J. Psychiatry 177, 38–41.

Arseneault, L., Cannon, M., Poulton, R., Murray, R., Caspi, A.,Moffitt, T.E., 2002. Cannabis use in adolescence and risk for adultpsychosis: longitudinal prospective study. Br. Med. J. 325,1212–1213.

Arseneault, L., Cannon, M., Witton, J., Murray, R.M., 2004. Causalassociation between cannabis and psychosis: examination of theevidence. Br. J. Psychiatry 184, 110–117.

Box, G.E.P., Jenkins, G.M., 1970. Time Series Analysis: Forecastingand Control. Holden-Day, San Francisco.

Boydell, J., Van Os, J., Lambri, M., Castle, D., Allardyce, J.,McCreadie, R.G., Murray, R.M., 2003. Incidence of schizophreniain south-east London between 1965 and 1997. Br. J. Psychiatry182, 45–49.

Boydell, J., van Os, J., Caspi, A., Kennedy, N., Giouroukou, E.,Fearon, P., Farrell, M., Murray, R.M., 2006. Trends in cannabis useprior to first presentation with schizophrenia, in South-EastLondon between 1965 and 1999. Psychol. Med. 36, 1441–1446.

Brewin, J., Cantwell, R., Dalkin, T., Fox, R., Medley, I., Glazebrook, C.,Kwiecinski, R., Harrison, G., 1997. Incidence of schizophrenia inNottingham. A comparison of two cohorts, 1978–80 and 1992–94.Br. J. Psychiatry 171, 140–144.

Caspi, A.,Moffitt, T.E., Cannon,M.,McClay, J., Murray, R., Harrington,H., Taylor, A., Arseneault, L.,Williams, B., Braithwaite, A., Poulton,R., Craig, I.W., 2005. Moderation of the effect of adolescent-onset

cannabis use on adult psychosis by a functional polymorphism in thecatechol-O-methyltransferase gene: longitudinal evidence of a geneX environment interaction. Biol. Psychiatry 57, 1117–1127.

Clayton, D., Schifflers, E., 1987a.Models for temporal variation in cancerrates, I: age-period and age-cohort models. Stat. Med. 6, 449–467.

Clayton, D., Schifflers, E., 1987b. Models for temporal variation incancer rates, II: age-period-cohort models. Stat. Med. 6, 469–481.

Daumann, J., Hensen, G., Thimm, B., Rezk, M., Till, B., Gouzoulis-Mayfrank, E., 2004. Self-reported psychopathological symptomsin recreational ecstasy (MDMA) users are mainly associated withregular cannabis use: further evidence from a combined cross-sectional/longitudinal investigation. Psychopharmacology (Berl)173, 398–404.

Degenhardt, L., Hall, W., Lynskey, M., 2003. Testing hypotheses aboutthe relationship between cannabis use and psychosis. DrugAlcohol Depend. 71, 37–48.

Delgrande Jordan, M., Kuntsche, E., Schmid, H., Kuendig, H., 2004.Trends im Substanzkonsum von Jugendlichen in der Schweiz[Trends in substance use pattern among adolescents in Switzer-land]. Paediatrica 15, 34–35.

De Preux, E., Dubois-Arber, F., Zobel, F., 2004. Current trends inillegal drug use and drug related health problems in Switzerland.Swiss Med. Wkly. 134, 313–321.

Der, G., Gupta, S., Murray, R.M., 1990. Is schizophrenia disappear-ing? Lancet 335, 513–516.

Drewe, M., Drewe, J., Riecher-Rossler, A., 2004. Cannabis and risk ofpsychosis. Swiss Med. Wkly. 134, 659–663.

Eagles, J.M., Whalley, L.J., 1985. Decline in the diagnosis ofschizophrenia among first admissions to Scottish mental hospitalsfrom 1969–78. Br. J. Psychiatry 146, 151–154.

Eagles, J.M., Hunter, D., McCance, C., 1988. Decline in the diagnosisof schizophrenia among first contacts with psychiatric services innorth-east Scotland, 1969–1984. Br. J. Psychiatry 152, 793–798.

Fergusson, D.M., Horwood, L.J., Swain-Campbell, N.R., 2003.Cannabis dependence and psychotic symptoms in young people.Psychol. Med. 33, 15–21.

Fergusson, D.M., Horwood, L.J., Ridder, E.M., 2005. Tests of causallinkages between cannabis use and psychotic symptoms. Addic-tion 100, 354–366.

Fienberg, S.E., Mason, W.M., 1985. Specification and implementation ofage, period and cohortmodels. In:Mason,W.M., Fienberg, S.E. (Eds.),Cohort Analysis in Social Research. Springer, New York, pp. 44–87.

Geddes, J.R., Black, R.J., Whalley, L.J., Eagles, J.M., 1993.Persistence of the decline in the diagnosis of schizophreniaamong first admissions to Scottish hospitals from 1969 to 1988. Br.J. Psychiatry 163, 620–626.

Gottman, J.M., 1981. Time-series analysis. Cambridge UniversityPress, Cambridge.

Hall, W., Degenhardt, L., Teesson, M., 2004. Cannabis use andpsychotic disorders: an update. Drug Alcohol Rev. 23, 433–443.

Harrison, G., Cooper, J.E., Gancarczyk, R., 1991. Changes in theadministrative incidence of schizophrenia. Br. J. Psychiatry 159,811–816.

Henquet, C., Krabbendam, L., Spauwen, J., Kaplan, C., Lieb, R.,Wittchen, H.U., van Os, J., 2005a. Prospective cohort study ofcannabis use, predisposition for psychosis, and psychotic symptomsin young people. Br. Med. J. 330, 11.

Henquet, C., Murray, R., Linszen, D., van Os, J., 2005b. Theenvironment and schizophrenia: the role of cannabis use. Schizophr.Bull. 31, 608–612.

Holford, T.R., 1985. An alternative approach to statistical age-period-cohort analysis. J. Chronic. Dis. 38, 831–836.


Holford, T.R., 1991. Understanding the effects of age, period, andcohort on incidence and mortality rates. Annu. Rev. Public Health12, 425–457.

Jones, I.H., Kirkby, K.C., Hay, D.A., Daniels, B.A., Longmore, L.M.,1997. Decline in diagnoses of schizophrenia in Tasmania duringthe period 1965–1990. Acta Psychiatr. Scand. 95, 13–18.

Kendell, R.E., Malcolm, D.E., Adams, W., 1993. The problem ofdetecting changes in the incidence of schizophrenia. Br. J. Psychiatry162, 212–218.

Kuntsche, E.N., 2004. Progression of a general substance use patternamong adolescents in Switzerland? Investigating the relationshipbetween alcohol, tobacco, and cannabis use over a 12-year period.Eur. Addict. Res. 10, 118–125.

Lauber, C., Lay, B., Rossler, W., 2005. Homelessness among peoplewith severe mental illness in Switzerland. Swiss Med. Wkly. 135,50–56.

Macleod, J., Oakes, R., Copello, A., Crome, I., Egger, M., Hickman, M.,Oppenkowski, T., Stokes-Lampard, H., Davey Smith, G., 2004.Psychological and social sequelae of cannabis and other illicit druguse by young people: a systematic review of longitudinal, generalpopulation studies. Lancet 363, 1579–1588.

Macleod, J., Davey Smith, G., Hickman, M., 2006. Does cannabis usecause schizophrenia? Lancet 367, 1055.

Maki, P., Veijola, J., Jones, P.B., Murray, G.K., Koponen, H., Tienari,P., Miettunen, J., Tanskanen, P., Wahlberg, K.E., Koskinen, J.,Lauronen, E., Isohanni, M., 2005. Predictors of schizophrenia— areview. Br. Med. Bull. 73, 1–15.

McCormick, S., Goff, D.C., 1991. Apparent decrease in schizophrenia.Br. J. Psychiatry 159, 725.

McGrath, J.J., 2005. Myths and plain truths about schizophreniaepidemiology — the NAPE lecture 2004. Acta Psychiatr. Scand.111, 4–11.

Munk-Jorgensen, P., Mortensen, P.B., 1992. Incidence and other aspectsof the epidemiology of schizophrenia in Denmark, 1971–87. Br. J.Psychiatry 161, 489–495.

Narring, F., Tschumper, A., Inderwildi Bonivento, L., Jeannin, A.,Addor, V., Bütikofer, A., Suris, J.C., Diserens, C., Alsaker, F.,Michaud, P.A., 2004. SMASH 2002: Swiss multicenter adolescentsurvey on health 2002. Raisons de santé. Institut universitaire demédecine sociale et préventive, Lausanne.

Oldehinkel, A.J., Giel, R., 1995. Time trends in the care-basedincidence of schizophrenia. Br. J. Psychiatry 167, 777–782.

Osby, U., Hammar, N., Brandt, L., Wicks, S., Thinsz, Z., Ekbom, A.,Sparen, P., 2001. Time trends in first admissions for schizophreniaand paranoid psychosis in Stockholm County, Sweden. Schizophr.Res. 47, 247–254.

Preti, A., Miotto, P., 2000. Increase in first admissions for schizophreniaand other major psychoses in Italy. Psychiatry Res. 94, 139–152.

Rey, J.M., Tennant, C.C., 2002. Cannabis and mental health. Br. Med.J. 325, 1183–1184.

SAS Institute Inc., 1996. SAS for Macintosh, Version 6.12. SASInstitute Inc., Cary, NC.

Suvisaari, J.M., Haukka, J.K., Tanskanen, A.J., Lonnqvist, J.K., 1999.Decline in the incidence of schizophrenia in Finnish cohorts bornfrom 1954 to 1965. Arch. Gen. Psychiatry 56, 733–740.

Takei, N., Lewis, G., Sham, P.C., Murray, R.M., 1996. Age-period-cohort analysis of the incidence of schizophrenia in Scotland.Psychol. Med. 26, 963–973.

Tsuchiya, K.J., Munk-Jorgensen, P., 2002. First-admission rates ofschizophrenia in Denmark, 1980–1997: have they been increas-ing? Schizophr. Res. 54, 187–191.

Van Kampen, J., Katz, M., 2001. Persistent psychosis after a singleingestion of ‘ecstasy’. Psychosomatics 42, 525–527.

van Os, J., 2004. Does the urban environment cause psychosis? Br. J.Psychiatry 184, 287–288.

van Os, J., Bak, M., Hanssen, M., Bijl, R.V., de Graaf, R., Verdoux, H.,2002. Cannabis use and psychosis: a longitudinal population-basedstudy. Am. J. Epidemiol. 156, 319–327.

von Sydow, K., Lieb, R., Pfister, H., Hofler, M., Wittchen, H.U., 2002.Use, abuse and dependence of ecstasy and related drugs inadolescents and young adults— a transient phenomenon? Resultsfrom a longitudinal community study. Drug Alcohol Depend. 66,147–159.

Waddington, J.L., Youssef, H.A., 1994. Evidence for a gender-specificdecline in the rate of schizophrenia in rural Ireland over a 50-yearperiod. Br. J. Psychiatry 164, 171–176.

Woogh, C., 2001. Is schizophrenia on the decline in Canada? Can. J.Psychiatry 46, 61–67.

Zammit, S., Lewis, G., 2004. Exploring the relationship betweencannabis use and psychosis. Addiction 99, 1353–1355.

Documents

Changing incidence of psychotic disorders among the young in Zurich